In 18 anaesthetized and artificially ventilated pigs, oleic acid was infused intravenously in order to induce a lung injury characterized by increased lung water content, decreased compliance and a ventilation/perfusion disturbance. After a stabilizing period, half of the animals (group P) underwent repeated plasmapheresis, which halved their plasma oncotic pressure (POP). The rest of the animals were bled and re-transfused with the shed blood, thus serving as a control group. In both groups, care was taken to keep the mean left atrial pressure as constant as possible. During plasmapheresis and "shamapheresis", there was no significant increase in venous admixture (F102 0.21 and 0.6) in either of the groups. At the end of the study, end-inspiratory pressure, dead space/tidal volume ratio and wet/dry lung weight ratio (WW/DW) were significantly higher in group P. Venous admixture and WW/DW correlated significantly with pulmonary arterial pressure and calculated pulmonary capillary pressure, but not with POP or POP minus pulmonary arterial occlusion pressure. It is concluded that reduction of plasma oncotic pressure may increase the lung water content in previously injured lungs, but this extra water accumulation does not necessarily impair oxygenation in the lungs.